Kaynak atama teknikleri ile büyük ölçekli dağıtımlı sistemlerde enerji maliyet optimizasyonu

Large scale distributed systems require massive amount of computing power to provide reliable services and to solve computationally complex problems. In that regard, energy needs in these systems are increasing rapidly and that brings substantial costs. Thus, reducing energy costs of such organizations is crucial to advance in these fields. In addition to explicitly reducing energy consumption, it is also possible to cut down the total electricity bill by exploiting spatial and temporal variations in electricity prices. In the first part of the thesis, we propose a volunteer computing network where peers can set monetary budgets, limiting the financial burden incurred on them due the usage of their computational resources. Assuming that the price of the electricity consumed by the peers has temporal variation, we show that our approach leads to an interesting task allocation problem, where the goal is to maximize the amount of work done by the peers without violating the monetary budget constraints set by the peers. We propose various polynomial time heuristic algorithms to the problem, which is NP-hard, and our extensive simulations show that our approach can increase the total amount of work done up to 35% compared to an existing baseline. In the second part, we consider a geographically distributed data center network that is specialized to run batch jobs with previously determined Service Level Agreements (SLAs). Taking into account the spatial and temporal variations in the electricity prices and free cooling opportunities by utilizing the outside weather, we model the problem of minimizing the energy cost as a linear programming problem. We propose two job scheduling heuristic algorithms and our simulations using real-life workload traces and electricity prices demonstrate that the proposed heuristics can decrease the total energy cost up to 6% compared to a load balancing baseline solution.